# AI tools that do one job are a dead end. We call them point solutions. In roofing, they are a dead end for the roofers who buy them, the companies that build them and the investors who fund them. Disclosure: Films and stills on this page are AI-generated illustrations. They show an idea, not any company, product, person or place, and they are not evidence. The 3D figures stand for counts, not real people or real companies. Every statement comes from Foundation's research paper "A shelf of tools is not a foundation" (October 2026), which lists the source and limits of each one and grades the 24 claims of the video it started from. Roofing figures are U.S. Census Bureau data for 2022 and 2023. Per-tool figures are multiplication at the paper's starting values, with an assumed eight tools: not a forecast and not a model of any real vendor. The Harvey demo is Foundation's first-hand account of one conversation. Foundation Projects builds a platform for roofing companies, so it is not neutral. This page is research and opinion. It is not investment, legal or insurance advice, and it is not an offer of securities. Music: generated with Higgsfield Sonilo. ## A working explanation ## Shared opening ### AI tools that do one job are a dead end. Here is what held up from the video that said so, and a test you can run on any tool. Picture: [Opening film: generated illustration · mood only] ### We checked a video: “Vertical AI is dead.” We checked “Vertical AI is DEAD,” a video by Devin Kearns. He posted it in September 2026; it runs 15 minutes and covers AI tools built for one industry. We turned it into 24 claims. Then we checked each one against primary sources. 4 held. 12 held in part. 8 could not be sourced. None was contradicted. This page uses only what held, in full or in part, in its corrected form. Picture: [Truck film: generated illustration · mood only] ### The playbook: find one expensive task. Add AI. For three years, a common AI startup pitch had one move. Find one costly task. Add AI. Sell the savings. In roofing, the tools pitched to us most are an AI phone receptionist, an insurance supplement writer and an instant-quote tool. They are real tasks with real money behind them. Some of them work. A narrow tool is an honest way into an industry. The question is what happens next. Picture: [Playbook film: generated illustration · mood only] ### ChatGPT and Claude keep taking in these tasks. ChatGPT and Claude keep taking in what these tools sell. The dates are public. 2023: ChatGPT searches your PDFs, and anyone can build a custom assistant without code. 2024: Anthropic launches an open standard to connect assistants to business systems. 2025: OpenAI's voice model can answer phone calls. Claude gets skills for spreadsheets, slides and documents. By October 2026: business tools plug straight into ChatGPT. Tools do not vanish. But what made them special is now easy to get. Picture: [Light film: generated illustration · mood only] ### Software gets a thin slice. Software gets a thin slice. In 2022, software and computer services were 0.33% of U.S. roofing receipts. Payroll was about 87 times the software line. People are the cost; software is a sliver. The average roofing company with employees took in about $2.9 million. Split across an assumed eight tools, one tool gets about $1,200 a year. Picture: [Paper stack film: generated illustration · mood only] ### One tool needs 35 of every 100 roofing companies. Each figure is 1 in 100 roofing companies with employees: about 24,000 in all. To earn $1 million a year, one tool needs 836 of them: about 3 in 100. To earn $10 million a year, it needs 8,360: 35 in 100. $100 million would take 3.5 times the number of roofing companies with employees. That is why each of these tools is pitched as the first step into something wider. - 3D scene (people-field): Roofing companies with employees, out of every 100. Each figure is 1 of 100 roofing companies. Each figure stands for 1 in 100 U.S. roofing companies with employees (24,082 in 2022). Not real people. - of every 100 roofing companies with employees would have to buy one tool for its maker to earn $1 million a year: 97 of 100 roofing companies with employees, 3 of 100 would have to buy one tool (exact value: 3.5 per 100). - of every 100 roofing companies with employees would have to buy one tool for its maker to earn $10 million a year: 65 of 100 roofing companies with employees, 35 of 100 would have to buy one tool (exact value: 34.7 per 100). - Arithmetic on an assumed eight tools, not a forecast ### Blank card: Shared opening - AI tools that do one job are a dead end.: The claim: AI point solutions are a dead end. - We checked a video: “Vertical AI is dead.”: The source: Devin Kearns, "Vertical AI is DEAD" (September 2026), checked claim by claim. - The playbook: find one expensive task. Add AI.: The roofing tools pitched most: AI phone receptionist, insurance supplement writer, instant-quote tool. - ChatGPT and Claude keep taking in these tasks.: Absorbed: what a tool was first valued for stops being scarce. - Software gets a thin slice.: The slice: software and computer services are 0.33% of roofing receipts; about $1,200 a year per tool (eight tools assumed). - One tool needs 35 of every 100 roofing companies.: $10 million a year from one tool = 35 of every 100 roofing companies as customers. ## Which one are you? The same problem ends in a different dead end for each of you. - Roofing company owners (You run a roofing company): What does a shelf of tools do to my company? See "The owner's path". - Software builders (You build software for roofers or other trades): Does my product have a moat? See "The builder's path". - Investors (You invest in software or services): Is an AI tool that does one roofing job a good bet? See "The investor's path". ## The owner's path ### One job. Nine steps. A tool for each. One roof job has nine steps. Follow it from the first call to the last review. Lead, inspection, estimate, claim, contract, production, materials, invoice, review: nine steps. In the point-solution pitch, each step gets its own tool. Each tool keeps its own copy of the homeowner. Picture: [Office film: generated illustration · mood only] ### Nine tools means you wire them together. The gaps between tools are handoffs. Nine tools, nine copies of the same homeowner. Buy a tool for every step, and you become the person who wires them together. Picture: [Corkboard film: generated illustration · mood only] ### If the tool gets it wrong, you are still responsible. If the tool gets it wrong, you are still responsible. An AI receptionist that answers a call can record it. Florida needs every party's consent to record a call. Texas needs one party's. Claim rules stay with you too. Texas and Florida decide who may act for a policyholder, and neither statute mentions software. In our reading, the duty to tell callers they are talking to AI lands on you, whichever tool takes the call. OpenAI tells developers to make it clear, unless it is already obvious. Picture: [Office film: generated illustration · mood only] ### Blank card: My tool check (owner) The four tests and the four questions, in our view. - One job. Nine steps. A tool for each.: No foundation: a tool for every step is not a system. - Nine tools means you wire them together.: No foundation: buy a tool per step and you become the integrator. - If the tool gets it wrong, you are still responsible.: The duty stays with you, whatever tool answers. - The tool I will put to the test first: #### The four tests (our view) - [ ] Data the assistants cannot get - [ ] A place in your workflow that is expensive to leave - [ ] A rule that requires it - [ ] An outcome it stands behind #### Ask before you buy - [ ] Does it write to my job record, or keep its own? - [ ] What happens to my data and workflows if I leave? - [ ] Who answers when it is wrong, and what does it cost me? - [ ] What does it do that my assistant will not do next year? #### Sources Every line comes from the research paper "A shelf of tools is not a foundation": foundationproject.net/research/ai-point-solutions-roofing. Its sources and the graded claim ledger are listed there. #### Foundation's interest Foundation Projects builds a platform for roofing companies, so we are not neutral. This is research and opinion, not investment, legal or insurance advice, and not an offer of securities. ## The builder's path ### Roofing AI tools get squeezed from both sides. A roofing AI tool gets squeezed from both sides. Above it, the assistants ship the skill: reading files, taking calls, connecting systems. Below it, the data owners keep their data. Verisk's license says: “We own all Price Data provided with the Services.” EagleView says it owns its aerial imagery. In April 2026 it announced its own AI product. In between sits the tool: a workflow and a screen. In our view the gap is packaging, and packaging is the easiest part to copy. Picture: [Slab film: generated illustration · mood only] ### The industry label is only where it started. The industry label is only where a company started. Harvey started with law firms; it now reports more than 1,000 customers in 60 countries. In 2025, corporate customers went from about 4% to 33% of Harvey's revenue, its chief executive told TechCrunch. Our first-hand account: in October 2026 Harvey demoed to us, a roofing platform, not a law firm. The offer: about $20,000 a year, no usage charges. The video said Harvey is becoming a law firm. We found no source for that. What we saw: a product sold as legal AI, offered to an outsider on software-license terms. Picture: [Meeting room film: generated illustration · mood only] The picture is a generated illustration, not Harvey's product or the demo. ### After the first sale, you need a moat. After the first sale, you need a moat. In 2023 Sequoia warned that data moats are “on shaky ground” and saw workflows and user networks lasting. After the introduction, a vendor needs one of three things: a place in the workflow that is expensive to leave, a network, or an outcome it stands behind. A tool with none of them is the first thing the next assistant release makes optional. A system the owner runs the whole business on is different. It is not a one-job tool; in our view it is a candidate foundation. Picture: [Tide film: generated illustration · mood only] ### Blank card: My moat check (builder) The four tests, in our view. - Roofing AI tools get squeezed from both sides.: No moat: the gap is packaging, and packaging is easy to copy. - The industry label is only where it started.: The vertical was a label: it says where a vendor started, not where it will sell. - After the first sale, you need a moat.: After the first sale: a costly-to-leave workflow, a network, or an outcome you stand behind. - Which of the four tests my product passes: #### The four tests (our view) - [ ] Data the assistants cannot get - [ ] A place in your workflow that is expensive to leave - [ ] A rule that requires it - [ ] An outcome it stands behind #### After the first sale The first sale buys an introduction. Then a vendor needs a place in the workflow that is expensive to leave, a network, or an outcome it stands behind. #### Sources Every line comes from the research paper "A shelf of tools is not a foundation": foundationproject.net/research/ai-point-solutions-roofing. Its sources and the graded claim ledger are listed there. #### Foundation's interest Foundation Projects builds a platform for roofing companies, so we are not neutral. This is research and opinion, not investment, legal or insurance advice, and not an offer of securities. ## The investor's path ### A seat tops out at the software budget. A seat tops out at the software budget. A tool sold per seat can only earn what a company spends on software. You saw it in the count: on the measured share, $100 million from one tool takes more customers than there are roofing companies with employees. Harvey's chief executive, November 2025: “Right now it’s mostly seats, but we’re moving to more outcome-based pricing as the workflows get more complex.” Picture: [Tide film: generated illustration · mood only] ### At 2%, $100 million needs 57 in 100. Even six times the budget leaves the ceiling low. Say roofers spent 2% of revenue on software, six times the measured share: about $7,250 a year for each of eight tools at an average company. $10 million a year would then take 1,380 customers: about 6 of every 100 roofing companies with employees. $100 million would take 13,794: 57 of every 100. Venture-scale revenue from one roofing tool needs most of the industry on generous assumptions, and more than all of it on measured ones. - 3D scene (people-field): Roofing companies with employees, out of every 100, if software were 2% of revenue. Each figure is 1 of 100 roofing companies. Each figure stands for 1 in 100 U.S. roofing companies with employees (24,082 in 2022). Not real people. - of every 100 roofing companies with employees would have to buy one tool for its maker to earn $10 million a year, if software were 2% of revenue: 94 of 100 roofing companies with employees, 6 of 100 would have to buy one tool (exact value: 5.7 per 100). - of every 100 roofing companies with employees would have to buy one tool for its maker to earn $100 million a year, if software were 2% of revenue: 43 of 100 roofing companies with employees, 57 of 100 would have to buy one tool (exact value: 57.3 per 100). - Arithmetic on an assumed 2% share and eight tools, not a forecast ### Some big investors are buying service companies instead. Some big investors are buying service companies instead of tools. General Catalyst raised $1.5 billion in 2024 to own and run them, PitchBook reported. Long Lake, an AI holding company it backs, closed its $6.3 billion purchase of American Express Global Business Travel on 29 September 2026. None of that proves the strategy works. Closing a purchase is not showing the model pays. The question is not whether the tools work. It is who keeps what they create. Picture: [Yard film: generated illustration · mood only] The picture is a generated illustration, not any company named here. ### Blank card: My pitch check (investor) The four tests and the question to ask, in our view. - A seat tops out at the software budget.: A low ceiling: a seat tops out at the software budget. - At 2%, $100 million needs 57 in 100.: Even at 2% of revenue, $100 million from one tool takes 57 of every 100 roofing companies with employees. - Some big investors are buying service companies instead.: The investor's question: who keeps what the tools create? - The question I will ask the next pitch: #### The four tests (our view) - [ ] Data the assistants cannot get - [ ] A place in your workflow that is expensive to leave - [ ] A rule that requires it - [ ] An outcome it stands behind #### The ceiling Software and computer services are 0.33% of roofing receipts (software alone 0.22%): about $9,570 a year for an average company. Split across eight tools (an assumption), about $1,200 each; $10 million a year then takes 35 of every 100 roofing companies as customers. #### Even at 2% If roofers spent 2% of revenue on software, six times the measured share, an average company would spend about $7,250 a year on each of eight tools (an assumption). One tool would then need about 6 of every 100 roofing companies with employees as customers to reach $10 million a year, and 57 of every 100 to reach $100 million. #### Sources Every line comes from the research paper "A shelf of tools is not a foundation": foundationproject.net/research/ai-point-solutions-roofing. Its sources and the graded claim ledger are listed there. #### Foundation's interest Foundation Projects builds a platform for roofing companies, so we are not neutral. This is research and opinion, not investment, legal or insurance advice, and not an offer of securities. ## Shared close ### The video's big point held. Some details did not. The video's big point held. Some details did not. Here are nine of its 24 claims, graded the same way: held, held in part, could not be sourced. Held: the big assistants already do much of what one-job tools sold. File search, skills and plugins shipped inside them, and a shared standard was launched to replace one-off connectors. Held: owning the service, not selling software to it, reaches the budget software never touches. General Catalyst, an investor that buys and runs service companies, says so and is funding it. Held in part: buyers want one foundation, not a shelf of tools. We found no buyer survey. Held in part: the claim that each era of AI tools makes the last one obsolete. Too strong: plugins are a ChatGPT feature again, and workflow automation is still sold. Held in part: the claim that software is about 2% of a law firm's revenue. Clio measures just under 2% of expenses, not revenue. In roofing, software alone is about a fifth of 1%. Could not be sourced: the claim that Harvey, a legal AI company, is becoming a law firm. Harvey says its commitment to law firms is not changing. Also among the 8 we could not source: Nvidia undercutting the AI labs, investors not knowing the ground, and distribution not being a moat. Picture: [Desk film: generated illustration · mood only] ### Put a tool to the test. A tool survives if it has something an assistant cannot copy. Four things count. - Put a tool to the test (tool-test). The four tests: - 01 Data the assistants cannot get: Does it hold data the assistants cannot get? - 02 A place in your workflow that is expensive to leave: Is it where you run the business, so it would be costly to leave? - 03 A rule that requires it: Does a law or rule require this tool, not just a licensed person? - 04 An outcome it stands behind: Is the vendor paid only when you get the result? - The rule: a tool stays when it has at least 1 of the four. Otherwise it falls. “Ask the vendor” never counts as yes. - A foundation, a stack, top to bottom: An operating layer, All the workflows, The data. Owners and crews: handle what the system cannot. Below the floor: Off the edge. - Tool: AI phone receptionist. Result: 0 of 4: falls. It falls. AI phone receptionist: we found none of the four. The next assistant release, a data owner or the system you run the business on can make it unnecessary. - Data the assistants cannot get: Not that we found. Phone calling is built into OpenAI's voice platform: calls since August 2025, $0.05 a minute for the voice session since September 2026, model and tool use extra. - A place in your workflow that is expensive to leave: Ask the vendor. Only if it is part of the system where you run the business. Jobber already sells an AI receptionist inside its field-service suite. - A rule that requires it: Not that we found. In our reading, the duty to tell callers they are talking to AI lands on whoever runs the line, whatever tool answers. - An outcome it stands behind: Ask the vendor. Is it paid on your result, or a flat fee per user? - Brief contribution: AI phone receptionist: it falls. We found none of the four. - Tool: Insurance supplement writer. Result: 0 of 4: falls. It falls. Insurance supplement writer: we found none of the four. The next assistant release, a data owner or the system you run the business on can make it unnecessary. - Data the assistants cannot get: Not that we found. The price data is Verisk's: “We own all Price Data provided with the Services.” Verisk now brings its own tools into Claude. - A place in your workflow that is expensive to leave: Ask the vendor. Only if it is part of the system where you run the business. Ask whether it keeps its own copy of your customers. - A rule that requires it: Not that we found. Texas and Florida decide who may act for a policyholder on a claim. In our reading, those rules protect the licensed adjuster, not the software that drafts the letter. - An outcome it stands behind: Ask the vendor. Is it paid on your result, or a flat fee per user? - Brief contribution: Insurance supplement writer: it falls. We found none of the four. - Tool: Instant-quote tool. Result: 0 of 4: falls. It falls. Instant-quote tool: we found none of the four. The next assistant release, a data owner or the system you run the business on can make it unnecessary. - Data the assistants cannot get: Not that we found. EagleView says it owns its aerial imagery and announced its own AI product in April 2026. Google's Solar API terms list energy uses, not roofing quotes. - A place in your workflow that is expensive to leave: Ask the vendor. Only if it is part of the system where you run the business. Ask whether it keeps its own copy of your customers. - A rule that requires it: Not that we found. In our reading, the Texas and Florida claim rules protect licensed people, not software. - An outcome it stands behind: Ask the vendor. Is it paid on your result, or a flat fee per user? - Brief contribution: Instant-quote tool: it falls. We found none of the four. - Tool: The data owner's own AI (Verisk, EagleView). Result: 1 of 4: stays. It stays. The data owner's own AI (Verisk, EagleView): 1 of the four. It has something an assistant cannot copy, so the argument does not apply. - Data the assistants cannot get: Yes. In our view, a tool built on data its maker owns is not a wrapper. Verisk says it owns its price data; EagleView says it owns its aerial imagery. Both have announced their own AI. - A place in your workflow that is expensive to leave: Ask the vendor. Only if it is part of the system where you run the business. Ask whether it keeps its own copy of your customers. - A rule that requires it: Not that we found. In our reading, the Texas and Florida claim rules protect licensed people, not software. - An outcome it stands behind: Ask the vendor. Is it paid on your result, or a flat fee per user? - Brief contribution: The data owner's own AI (Verisk, EagleView): it stays. It passes 1 of the four tests. - Tool: Your job record (CRM and field system). Result: 1 of 4: stays. It stays. Your job record (CRM and field system): 1 of the four. It has something an assistant cannot copy, so the argument does not apply. - Data the assistants cannot get: Ask the vendor. It holds your own jobs and customers. Ask what happens to that data if you leave. - A place in your workflow that is expensive to leave: Yes. In our reading, the system where you run the business is costly to leave. Sequoia wrote in 2023 that workflows hold up better than data. - A rule that requires it: Not that we found. In our reading, the Texas and Florida claim rules protect licensed people, not software. - An outcome it stands behind: Ask the vendor. Does the vendor stand behind a result? Contractor systems like Jobber and Contractors Cloud are already opening to the assistants. - Brief contribution: Your job record (CRM and field system): it stays. It passes 1 of the four tests. - A tool you use, build or back: Answer the four questions for it. - Does it hold data the assistants cannot get? (Yes / No / Not sure) - Is it where you run the business, so it would be costly to leave? (Yes / No / Not sure) - Does a law or rule require this tool, not just a licensed person? (Yes / No / Not sure) - Is the vendor paid only when you get the result? (Yes / No / Not sure) - Count the “Yes” answers. 1 or more of 4: It stays. Fewer: It falls. - You answered “Not sure” at least once. Ask the vendor before you buy, renew or invest. - Send this test to someone (after a verdict): the link names only the tool and its result (“it fell” or “it stayed”), never your answers. Opened from it, the test says “A friend sent you their test. AI phone receptionist: it fell. Put your own tool to the test.” and replays that tool's verdict. ### A one-job tool becomes someone else's feature. A one-job tool becomes someone else's feature. The cause is the same for everyone: the budget is thin, and the big assistants take in the task. The budget is thin: split across an assumed eight tools, one tool gets about $1,200 a year from a roofing company. Picture: [Falling cards film: generated illustration · mood only] ### What lasts: doing the job and standing behind it. What lasts: doing the job and standing behind it. In our view, three things gain value as AI improves: how a company runs its jobs, the records they leave, and standing behind the result. Own all three and stand behind the result. We call that a foundation. When the system gets stuck, owners and crews take over. About 215,000 people work at U.S. roofing companies with employees. A tool gets a small cut of each job. The company that does the job keeps the rest: how it runs, its job records and its name on the result. Picture: [Sunrise film: generated illustration · mood only] The street is a generated illustration, not a photo of any company or crew. - 3D scene (sunrise-crowd): People at U.S. roofing companies with employees. 1 figure = about 2,000 people. Measured · Census County Business Patterns 2023. Each figure stands for about 2,000 people. Not real people. - people working at U.S. roofing companies with employees (2023): 108 figures (about 215,242 people). - You stand in front of the crowd and are not counted. ### Your next step. Every claim on this page comes from the research paper, which lists each source and its limits. Picture: [Closing film: generated illustration · mood only] - Calls to action: - Roofing company owners: Start a private conversation (For roofing company owners: talk through your tools and your system with us.): https://foundationproject.net/schedule; Save my results (opens the plan card) - Software builders: Start a private conversation (Building for roofers? Compare notes with us.): https://foundationproject.net/schedule; Save my results (opens the plan card) - Investors: Join the investor waitlist (The current round is closed. The waitlist sends one email if another round opens.): https://foundationproject.net/investor-packet; Save my results (opens the plan card) - Read the research paper: https://foundationproject.net/research/ai-point-solutions-roofing - Watch the video we checked: https://youtu.be/Wc4C7iz3cA0 - Download the claim ledger (CSV): https://foundationproject.net/research/point-solutions/claim-ledger.csv - Disclosure: Films on this page are AI-generated illustrations, not footage of any company or product. Roofing figures are U.S. Census data, per-tool figures are arithmetic on an assumed eight tools (not a forecast), and Foundation, which builds a platform for roofing companies, is not neutral. ### Blank card: Shared close - The video's big point held. Some details did not.: Lessons from the video that held, with their grades. - Put a tool to the test.: The tool I tested and how many of the four tests it passed. - A one-job tool becomes someone else's feature.: One cause: the budget is thin and the big assistants take in the task. - What lasts: doing the job and standing behind it.: What lasts: workflows, the data they leave behind, and who answers for the result. - Your next step.: Next step. Put a tool to the test: Tool I tested: ________ · Tests it has: __ of 4 Tests it has (tick each): [ ] Data the assistants cannot get · [ ] A place in your workflow that is expensive to leave · [ ] A rule that requires it · [ ] An outcome it stands behind ## Claims, sources and limits Review is an editorial status. A reviewed claim is not independent proof of a measured outcome. ### cl-thesis / editorial / reviewed Label: Our reading AI point solutions for roofers are a dead end: for the software companies that build them, the roofers who buy them and the investors who fund them. Scope: The paper's thesis Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 17 - Selected supporting excerpt: AI point solutions for roofers are a dead end: for the software companies that build them, the roofers who buy them, and the investors who fund them. - Limits: As written in the paper. ### cl-three / editorial / reviewed Label: Our reading For the software company: no moat. For the roofer: no foundation. For the investor: a low ceiling. Scope: The paper's short version Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 29 - Selected supporting excerpt: **No moat.** - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 33 - Selected supporting excerpt: **No foundation.** - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 37 - Selected supporting excerpt: **A low ceiling.** - Limits: As written in the paper. ### cl-stake / editorial / reviewed Label: Foundation position Foundation Projects is building a founder-aligned platform for roofing companies, so it is not neutral on this question. Scope: Foundation's stake Limitations: Stated by Foundation about itself. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 43 - Selected supporting excerpt: Foundation Projects is building a founder-aligned platform for roofing companies, so we are not neutral on this question. - Limits: As written in the paper. ### cl-kearns / documented / reviewed Label: Attribution The argument began as a 15-minute video essay by Devin Kearns, "Vertical AI is DEAD", published 30 September 2026. The paper cites it as framing, not as evidence. Scope: The source of the framing Limitations: A commentator's argument and predictions, not evidence. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 314 - Selected supporting excerpt: The argument began as a 15-minute video essay by Devin Kearns - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 314 - Selected supporting excerpt: published 30 September 2026. We cite it as the source of the framing, not as evidence. - Limits: As written in the paper. ### cl-audit / editorial / reviewed Label: Our reading Foundation turned the video into 24 checkable claims and graded each against primary sources: 4 held as stated, 12 held in part, 8 could not be sourced and none was contradicted. Scope: Foundation's claim audit, published as a ledger Limitations: Foundation's own grading; an independent fact-check moved four grades down before publication. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 41 - Selected supporting excerpt: We turned it into 24 checkable claims and graded each one against primary sources. 4 held as stated, 12 held in part, 8 could not be sourced, and none was contradicted outright. - Limits: As written in the paper. ### cl-audit-use / editorial / reviewed Label: Project decision The paper uses supported claims, and partial claims only in their corrected form. Unsupported claims are not used. Scope: How the audit is used Limitations: This page follows the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 334 - Selected supporting excerpt: The paper uses supported claims, and partial claims only in their corrected form. Unsupported and contradicted claims are not used. - Limits: As written in the paper. ### cl-disclosure / editorial / reviewed Label: Foundation position This is research and opinion from an interested party. It is not investment, legal or insurance advice, and it is not an offer of securities. Scope: Disclosure Limitations: Stated by Foundation about itself. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 304 - Selected supporting excerpt: This paper is research and opinion from an interested party. It is not investment, legal or insurance advice, and it is not an offer of securities. - Limits: As written in the paper. ### cl-playbook / editorial / reviewed Label: Our reading For three years a common AI startup pitch ran on one move: find one expensive, well-defined workflow, add AI, sell the return. Scope: The one-task playbook Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 47 - Selected supporting excerpt: For three years a common AI startup pitch has run on one move. Find one expensive, well-defined workflow. Add AI. Sell the return. - Limits: As written in the paper. ### cl-wedge-honest / editorial / reviewed Label: Our reading While models could not do much alone, a narrow tool with a clear payback was an honest way into an industry. Scope: The one-task playbook Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 47 - Selected supporting excerpt: a narrow tool with a clear payback was an honest way into an industry. - Limits: As written in the paper. ### cl-wedges / editorial / reviewed Label: Our reading The one-task roofing tools pitched to Foundation most often are an AI voice receptionist, an insurance supplement writer and an instant-quote tool. They are real tasks with real money behind them, and some of them work. Scope: Foundation's own experience Limitations: Foundation's experience, not a survey. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 49 - Selected supporting excerpt: In our own experience, the wedges pitched to us most often are an AI voice receptionist, an insurance supplement writer and an instant-quote tool. They are real tasks with real money behind them, and some of them work. - Limits: As written in the paper. ### cl-sequoia-2023 / documented / reviewed Label: Attribution In September 2023 Sequoia described the first wave of AI apps as copilots with undifferentiated pitches and terrible user retention. Scope: Sequoia Capital, "Generative AI's Act Two" (2023) Limitations: An investor's 2023 view. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 53 - Selected supporting excerpt: In September 2023 [Sequoia described the first wave] - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 53 - Selected supporting excerpt: every company was an “AI copilot,” its inbox was full of “undifferentiated pitches,” and user retention was “terrible.” - Limits: As written in the paper. ### cl-gpts / documented / reviewed Label: Official description November 2023: OpenAI's GPTs let anyone build custom assistants without code, and ChatGPT searches attached PDFs. Scope: OpenAI, "Introducing GPTs" (2023) Limitations: A launch post. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 61 - Selected supporting excerpt: November 2023 - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 65 - Selected supporting excerpt: Custom assistants from instructions, knowledge and skills. No code. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 67 - Selected supporting excerpt: You can also attach files to let ChatGPT search PDFs and other document types. - Limits: As written in the paper. ### cl-mcp / documented / reviewed Label: Official description November 2024: Anthropic launched the Model Context Protocol, one open protocol to connect assistants to business systems. Scope: Anthropic, "Introducing the Model Context Protocol" (2024) Limitations: A vendor announcement. Adoption is not measured. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 73 - Selected supporting excerpt: November 2024 - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 77 - Selected supporting excerpt: One open protocol to connect assistants to business systems. - Limits: As written in the paper. ### cl-skills / documented / reviewed Label: Official description October 2025: Anthropic shipped Agent Skills for Excel, PowerPoint, Word and fillable PDFs. Scope: Anthropic, "Introducing Agent Skills" (2025) Limitations: A vendor announcement. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 85 - Selected supporting excerpt: October 2025 - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 89 - Selected supporting excerpt: Shipped for Excel, PowerPoint, Word and fillable PDFs. - Limits: As written in the paper. ### cl-plugins / documented / reviewed Label: Official description By October 2026 OpenAI's plugins page invited business users to connect their apps "right into ChatGPT", from finance and sales to customer support and small business. Scope: OpenAI plugins page, read October 2026 Limitations: A product page with no launch date. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 97 - Selected supporting excerpt: By October 2026 - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 101 - Selected supporting excerpt: A page of business tools, from finance and sales to customer support and small business, to connect to ChatGPT and Codex. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 135 - Selected supporting excerpt: By October 2026 OpenAI’s plugins page was inviting business users to connect their apps “right into ChatGPT,” - Limits: As written in the paper. ### cl-scarce / editorial / reviewed Label: Our reading The timeline does not say every tool disappears. It says the thing a tool was first valued for stops being scarce. Scope: Reading the timeline Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 135 - Selected supporting excerpt: Read in order, the timeline does not say every tool disappears. It says the thing a tool was first valued for stops being scarce. - Limits: As written in the paper. ### cl-voice / documented / reviewed Label: Official description In August 2025 OpenAI's realtime voice model added phone calling over SIP. In September 2026 its newest voice model went generally available at $0.05 a minute for the voice session, with model and tool usage charged separately. Scope: OpenAI launch post, SIP guide and API changelog Limitations: Developer documentation; not a roofing product. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 139 - Selected supporting excerpt: In August 2025 [OpenAI’s realtime voice model added phone calling over SIP] - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 139 - Selected supporting excerpt: at $0.05 a minute for the voice session, billed per second, with model and tool usage charged separately. - Limits: As written in the paper. ### cl-verisk / documented / reviewed Label: Official description Verisk's license says "We own all Price Data provided with the Services", and in May 2026 it announced connectors that bring its analytics into Claude. Scope: Verisk license terms and announcement Limitations: Vendor terms and announcement; access requires Verisk subscriptions. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 141 - Selected supporting excerpt: say, “We own all Price Data provided with the Services.” In May 2026 it announced [connectors] - Limits: As written in the paper. ### cl-eagleview / documented / reviewed Label: Official description EagleView states that "All aerial imagery is collected and owned by Eagleview", and in April 2026 it announced its own agentic AI product. Scope: EagleView data disclosure and press release Limitations: Company statements; they describe EagleView's own data. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 143 - Selected supporting excerpt: EagleView states that [“All aerial imagery is collected and owned by Eagleview,”] - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 143 - Selected supporting excerpt: and in April 2026 it announced its [own agentic AI product] - Limits: As written in the paper. ### cl-solar / documented / reviewed Label: Official description Google's Solar API returns the dimensions of a building, but its terms list energy-system uses, not roofing quotes. Scope: Google Solar API documentation and terms Limitations: Terms as read in October 2026; not legal advice. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 143 - Selected supporting excerpt: but its [terms](https://cloud.google.com/maps-platform/terms/maps-service-terms) list energy-system uses, not roofing quotes. - Limits: As written in the paper. ### cl-packaging / editorial / reviewed Label: Our reading A roofing point solution sits between two owners: the assistants above ship the capability, the data owners below keep their data and ship their own AI. In Foundation's view the gap is packaging, and packaging is the easiest part to copy. Scope: Reading the three tools Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 145 - Selected supporting excerpt: Put the three together and a roofing point solution sits between two owners. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 145 - Selected supporting excerpt: In our view it is a gap in packaging, not in capability, and packaging is the easiest part to copy. - Limits: As written in the paper. ### cl-harvey-start / self-reported / reviewed Label: Vendor-reported Harvey's founders made "the explicit decision to focus our attention on law firms". Harvey now reports more than 1,000 customers in 60 countries. Scope: Harvey blog, March 2026 Limitations: Company-reported figures. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 151 - Selected supporting excerpt: made “the explicit decision to focus our attention on law firms.” - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 151 - Selected supporting excerpt: they report more than 1,000 customers in 60 countries - Limits: As written in the paper. ### cl-harvey-mix / documented / reviewed Label: Attribution Harvey's chief executive told TechCrunch that about 4% of revenue came from corporates at the start of 2025 and 33% did by November. Its March 2026 post also says "Nothing about our commitment to law firms is changing." Scope: TechCrunch (2025) and Harvey's blog (2026) Limitations: The executive's own figures. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 153 - Selected supporting excerpt: The same post says, “Nothing about our commitment to law firms is changing.” - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 153 - Selected supporting excerpt: about 4% of revenue came from corporates at the start of the year and 33% did by then. - Limits: As written in the paper. ### cl-harvey-demo / self-reported / reviewed Label: First-hand account In October 2026 Harvey demoed its product to Foundation, a roofing platform, not a law firm, and offered it for about $20,000 a year with no usage charges. Scope: Foundation's first-hand account Limitations: Foundation's first-hand account of one demo in October 2026. Not confirmed with Harvey; no public price list to check it against; terms vary by buyer. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 157 - Selected supporting excerpt: In October 2026 we sat through a Harvey demo. We are a roofing platform, not a law firm. Harvey offered to sell us the product anyway: about $20,000 a year, with no usage charges. - Limits: As written in the paper. ### cl-label / editorial / reviewed Label: Our reading This is not Harvey becoming a law firm. A vertical label describes where a vendor started. It does not describe where the vendor will sell. Scope: Reading the Harvey record Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. Foundation's first-hand account of one demo in October 2026. Not confirmed with Harvey; no public price list to check it against; terms vary by buyer. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 161 - Selected supporting excerpt: This is not Harvey becoming a law firm. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 161 - Selected supporting excerpt: A vertical label describes where a vendor started. It does not describe where the vendor will sell. - Limits: As written in the paper. ### cl-census / measured / reviewed Label: Official estimate In 2022 U.S. roofing contractors with employees reported $69.9 billion in receipts. They paid $13.5 billion in payroll and expensed $154 million of software: 0.22% of receipts, or 0.33% counting purchased computer services. Payroll was about 87 times the software line. Scope: Economic Census 2022, NAICS 238160 Limitations: U.S. Census Bureau, Economic Census 2022 and County Business Patterns 2023, NAICS 238160, as read in the paper. Employer firms only; 2022 data, before most roofing AI tools existed; software is expensed purchases only. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 169 - Selected supporting excerpt: roofing contractors with employees reported $69.9 billion in receipts. They paid $13.5 billion in payroll and expensed $154 million of software purchases, which is 0.22% of receipts, or 0.33% counting purchased computer services. Payroll was about 87 times the software line. - Limits: As written in the paper. ### cl-firms / measured / reviewed Label: Official estimate About 24,000 roofing firms in the U.S. have employees (24,082 in the 2022 Economic Census). Scope: Economic Census 2022, NAICS 238160 Limitations: U.S. Census Bureau, Economic Census 2022 and County Business Patterns 2023, NAICS 238160, as read in the paper. Employer firms only; 2022 data, before most roofing AI tools existed; software is expensed purchases only. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 254 - Selected supporting excerpt: About 24,000 roofing firms in the U.S. have employees - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 377 - Selected supporting excerpt: | Roofing firms with paid employees | 24,082 | 2022 | - Limits: As written in the paper. ### cl-employees / measured / reviewed Label: Official estimate About 215,000 people work at U.S. roofing establishments with paid employees (215,242 in the March 2023 pay period). Scope: County Business Patterns 2023, NAICS 238160 Limitations: U.S. Census Bureau, Economic Census 2022 and County Business Patterns 2023, NAICS 238160, as read in the paper. Employer firms only; 2022 data, before most roofing AI tools existed; software is expensed purchases only. Employment is the March 12 pay period; businesses with no employees are excluded. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 254 - Selected supporting excerpt: all of them together employ about 215,000 people - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 379 - Selected supporting excerpt: | Employees at those establishments (March 12 pay period) | 215,242 | 2023 | - Limits: As written in the paper. ### cl-average / inferred / reviewed Label: Derived arithmetic The average U.S. roofing firm with employees took in about $2.9 million in 2022. Scope: Derived from the Economic Census 2022 Limitations: U.S. Census Bureau, Economic Census 2022 and County Business Patterns 2023, NAICS 238160, as read in the paper. Employer firms only; 2022 data, before most roofing AI tools existed; software is expensed purchases only. Derived: receipts divided by firms. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 385 - Selected supporting excerpt: | Average receipts per roofing firm with employees | $2,902,840 | derived | - Limits: As written in the paper. ### cl-seat / inferred / reviewed Label: Derived arithmetic At the measured share, spread across an assumed eight tools, the average roofing company spends about $1,200 a year on any one of them ($1,196). A vendor selling one tool at that price needs 8,360 customers to reach $10 million a year, 35% of all roofing firms with employees, and 83,595 to reach $100 million, 3.5 times the number of roofing firms with employees. Scope: The paper's explorer at its starting values Limitations: Multiplication at the paper's starting values: $2.9 million revenue and 0.33% software share are measured averages; eight tools is an assumption, not a measurement. Not a forecast and not a model of any real vendor. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 201 - Selected supporting excerpt: At the measured share, spread across an assumed eight tools, the average roofing company spends $1,196 a year on any one of them. A vendor selling one tool at that price needs 8,360 customers to reach $10 million a year: 35% of all roofing firms with employees. To reach $100 million it needs 83,595: 3.5 times the number of roofing firms with employees. - Limits: As written in the paper. - seat-arithmetic / line 18 - Selected supporting excerpt: "customers_per_100_firms": 34.71, - Limits: Seat arithmetic file derived from the paper's explorer. ### cl-seat-1m / inferred / reviewed Label: Derived arithmetic At about $1,200 a year a tool, a vendor needs 836 roofing customers to earn $1 million a year: about 3 of every 100 roofing firms with employees. Scope: The paper's explorer method at its starting values Limitations: Multiplication at the paper's starting values: $2.9 million revenue and 0.33% software share are measured averages; eight tools is an assumption, not a measurement. Not a forecast and not a model of any real vendor. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 394 - Selected supporting excerpt: The customers a single-tool vendor needs are the revenue goal divided by one tool’s take, rounded up, set against the 24,082 roofing firms with employees. - Limits: As written in the paper. - seat-arithmetic / line 32 - Selected supporting excerpt: "customers_needed": 836, - Limits: Seat arithmetic file derived from the paper's explorer. - seat-arithmetic / line 34 - Selected supporting excerpt: "customers_per_100_firms": 3.47, - Limits: Seat arithmetic file derived from the paper's explorer. ### cl-ceiling / editorial / reviewed Label: Our reading A seat-based tool tops out at what a company spends on software. Scope: For the investor Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 37 - Selected supporting excerpt: A seat-based wedge tops out at what a company spends on software. - Limits: As written in the paper. ### cl-two-percent-test / inferred / reviewed Label: Derived arithmetic Even at 2%, six times the share measured with computer services, $10 million takes 5.7% of the roofing industry and $100 million takes 57% of it. Venture-scale revenue from one roofing tool needs most of the industry on generous assumptions and more than all of it on measured ones. Scope: The paper's explorer at an assumed 2% share Limitations: Multiplication at the paper's starting values: $2.9 million revenue and 0.33% software share are measured averages; eight tools is an assumption, not a measurement. Not a forecast and not a model of any real vendor. The 2% share is an assumption, six times the measured share. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 203 - Selected supporting excerpt: At 2%, six times the share measured with computer services, $10 million takes 5.7% of the industry and $100 million takes 57% of it. Venture-scale revenue from one roofing tool needs most of the industry on generous assumptions and more than all of it on measured ones. - Limits: As written in the paper. ### cl-two-percent-10m / inferred / reviewed Label: Derived arithmetic At 2% of revenue spread across an assumed eight tools, the average roofing company would spend about $7,250 a year on one tool. $10 million a year at that price takes 1,380 customers: about 6 of every 100 roofing firms with employees. Scope: The paper's explorer method at an assumed 2% share Limitations: Multiplication at the paper's starting values: $2.9 million revenue and 0.33% software share are measured averages; eight tools is an assumption, not a measurement. Not a forecast and not a model of any real vendor. The 2% share is an assumption, six times the measured share. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 394 - Selected supporting excerpt: The customers a single-tool vendor needs are the revenue goal divided by one tool’s take, rounded up, set against the 24,082 roofing firms with employees. - Limits: As written in the paper. - seat-arithmetic / line 41 - Selected supporting excerpt: "per_tool": 7250.0, - Limits: Seat arithmetic file derived from the paper's explorer at an assumed 2% share. - seat-arithmetic / line 42 - Selected supporting excerpt: "customers_needed": 1380, - Limits: Seat arithmetic file derived from the paper's explorer at an assumed 2% share. - seat-arithmetic / line 44 - Selected supporting excerpt: "customers_per_100_firms": 5.73, - Limits: Seat arithmetic file derived from the paper's explorer at an assumed 2% share. ### cl-two-percent-100m / inferred / reviewed Label: Derived arithmetic At the same $7,250 a tool, $100 million a year takes 13,794 customers: about 57 of every 100 roofing firms with employees. Scope: The paper's explorer method at an assumed 2% share Limitations: Multiplication at the paper's starting values: $2.9 million revenue and 0.33% software share are measured averages; eight tools is an assumption, not a measurement. Not a forecast and not a model of any real vendor. The 2% share is an assumption, six times the measured share. - seat-arithmetic / line 52 - Selected supporting excerpt: "customers_needed": 13794, - Limits: Seat arithmetic file derived from the paper's explorer at an assumed 2% share. - seat-arithmetic / line 54 - Selected supporting excerpt: "customers_per_100_firms": 57.28, - Limits: Seat arithmetic file derived from the paper's explorer at an assumed 2% share. ### cl-harvey-outcome / documented / reviewed Label: Attribution Harvey's chief executive told TechCrunch in November 2025: “Right now it’s mostly seats, but we’re moving to more outcome-based pricing as the workflows get more complex.” Scope: TechCrunch interview, November 2025 Limitations: The executive's own words. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 205 - Selected supporting excerpt: “Right now it’s mostly seats,” its chief executive told - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 205 - Selected supporting excerpt: “but we’re moving to more outcome-based pricing as the workflows get more complex.” - Limits: As written in the paper. ### cl-wider / editorial / reviewed Label: Our reading That is why each of these one-task tools is pitched as the first step into something wider. Scope: Reading the arithmetic Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 203 - Selected supporting excerpt: That is why every one of these wedges is pitched as the first step into something wider. - Limits: As written in the paper. ### cl-cause / editorial / reviewed Label: Our reading The cause is the same each time: the budget is thin and the assistants take in the task. What differs is who is left holding it. Scope: Three dead ends, one cause Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 209 - Selected supporting excerpt: The cause is the same each time: the seat is thin and the wedge gets absorbed. What differs is who is left holding it. - Limits: As written in the paper. ### cl-moats / documented / reviewed Label: Attribution Sequoia warned in 2023 that data moats are "on shaky ground" and saw workflows and user networks building more durable advantage. Scope: Sequoia Capital, "Generative AI's Act Two" (2023) Limitations: An investor's 2023 view. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 213 - Selected supporting excerpt: the ‘data moats’ are on shaky ground - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 213 - Selected supporting excerpt: “workflows and user networks.” - Limits: As written in the paper. ### cl-after-intro / editorial / reviewed Label: Our reading A one-task tool buys an introduction. After it, a vendor needs a place in the customer's workflow that is expensive to leave, a network around it, or an outcome it stands behind. A tool that holds none of them is the first thing the next assistant release makes optional. Scope: For the software company Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 217 - Selected supporting excerpt: So a wedge buys an introduction. What a vendor needs after the introduction is one of three things: a place in the customer’s workflow that is expensive to leave, a network around it, or an outcome it stands behind. A tool that does one step well and holds none of them is the first thing the next assistant release makes optional. - Limits: As written in the paper. ### cl-systems / documented / reviewed Label: Official description Contractor systems of record are opening to the assistants: Contractors Cloud publishes a guide for connecting Claude or ChatGPT through MCP, and Jobber, which sells an AI receptionist inside its field-service suite, opened its data to ChatGPT and Claude in September 2026. Scope: Contractors Cloud and Jobber pages Limitations: Vendor pages and releases. Jobber is a general field-service suite, not roofing-specific. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 219 - Selected supporting excerpt: [Contractors Cloud](https://api.contractorscloud.com/docs/guides/connect-your-ai-agent) publishes a guide for connecting Claude, ChatGPT or another AI agent to a customer’s account through MCP. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 219 - Selected supporting excerpt: Jobber, a general field-service suite, sells [an AI receptionist](https://www.getjobber.com/features/ai-receptionist/) inside its product and in September 2026 - Limits: As written in the paper. ### cl-sor / editorial / reviewed Label: Our reading A roofing system where the owner actually runs the business, the CRM and field system of record, is not a point solution. It is a candidate foundation. Scope: For the software company Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 219 - Selected supporting excerpt: A roofing system where the owner actually runs the business, the CRM and field system of record, is not a point solution in the sense this paper means. It is a candidate foundation - Limits: As written in the paper. ### cl-integrator / editorial / reviewed Label: Our reading Follow one job from the first call to the last review. In the point-solution pitch each step gets its own tool, and each tool keeps its own copy of the homeowner. An owner who buys a tool per step becomes the integrator. Scope: For the roofer Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 223 - Selected supporting excerpt: Follow one job from the first call to the last review. In the point-solution pitch each step gets its own tool, and each tool keeps its own copy of the homeowner. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 223 - Selected supporting excerpt: An owner who buys a tool per step becomes the integrator. - Limits: As written in the paper. ### cl-nine / editorial / reviewed Label: Our reading One roofing job passes through nine steps: lead, inspection, estimate, insurance claim, contract, production, materials, invoice and review. Scope: For the roofer Limitations: Illustrative, not a count of what any roofer uses. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 225 - Selected supporting excerpt: 1. Lead - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 233 - Selected supporting excerpt: 9. Review - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 237 - Selected supporting excerpt: One roofing job passes through nine steps, from the first call to the review. - Limits: As written in the paper. ### cl-recording / documented / reviewed Label: Official description Florida requires every party's consent to record a call; Texas requires one party's. Scope: Florida Statutes 934.03; Texas Penal Code 16.02 Limitations: Statute text, read October 2026. Not legal advice. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 243 - Selected supporting excerpt: requires every party’s consent](https://www.leg.state.fl.us - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 243 - Selected supporting excerpt: [Texas requires one party’s](https://tcss.legis.texas.gov/resources/PE/htm/PE.16.htm) - Limits: As written in the paper. ### cl-adjuster / documented / reviewed Label: Official description Texas bars a contractor from acting as a public adjuster on its own jobs, and Florida bars a licensed contractor from performing public adjuster services unless it is also licensed as one. Scope: Texas Insurance Code 4102; Florida Statutes 626.854 Limitations: Statute text, read October 2026. Not legal advice. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 243 - Selected supporting excerpt: [Texas bars a contractor from acting as a public adjuster](https://tcss.legis.texas.gov/resources/IN/htm/IN.4102.htm) on its own jobs - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 243 - Selected supporting excerpt: unless it is also licensed as one. - Limits: As written in the paper. ### cl-duty / editorial / reviewed Label: Our reading In Foundation's reading, the duty to tell callers they are talking to AI lands on whoever runs the phone line, whatever tool answers it. Scope: For the roofer Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. Not legal advice. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 243 - Selected supporting excerpt: “must also make it clear to end users when they’re interacting with AI, unless it’s already obvious from the context.” In our reading, that duty lands on whoever runs the phone line, whatever tool answers it. - Limits: As written in the paper. ### cl-buyer-questions / editorial / reviewed Label: Our reading Four questions to put to any tool before you buy it: Does it write to your job record, or keep its own? What happens to your data and your workflows if you leave? Who answers when it is wrong, and what does that cost you? What does it do that the assistant you already pay for will not do next year? Scope: For the roofer Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 247 - Selected supporting excerpt: Does it write to your job record, or keep its own? - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 248 - Selected supporting excerpt: What happens to your data and your workflows if you leave? - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 249 - Selected supporting excerpt: Who answers when it is wrong, and what does that cost you? - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 250 - Selected supporting excerpt: What does it do that the assistant you already pay for will not do next year? - Limits: As written in the paper. ### cl-gc / documented / reviewed Label: Attribution General Catalyst writes that technology builders capture more value by owning and operating the service businesses; it raised $1.5 billion in 2024 for that strategy, according to PitchBook. Scope: General Catalyst essay; PitchBook (2026) Limitations: An investor's thesis and secondary reporting. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 256 - Selected supporting excerpt: “technology builders capture even more value by directly owning and operating the service businesses - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 256 - Selected supporting excerpt: It raised $1.5 billion in 2024 for that strategy, according to [PitchBook] - Limits: As written in the paper. ### cl-longlake / self-reported / reviewed Label: Partner announcement Long Lake, an AI holding company backed by General Catalyst among others, completed its $6.3 billion purchase of American Express Global Business Travel on 29 September 2026. Scope: Company press release Limitations: A press release; it says nothing about returns or margins. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 258 - Selected supporting excerpt: Long Lake, an AI holding company backed by General Catalyst among others, [completed its $6.3 billion purchase of American Express Global Business Travel] - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 258 - Selected supporting excerpt: on 29 September 2026. - Limits: As written in the paper. ### cl-not-proof / editorial / reviewed Label: Our reading None of that proves the strategy works. Closing a purchase is not the same as showing the model pays. What the deals show is where the capital is pointed: at owning the service, not at selling it a seat. Scope: For the investor Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 260 - Selected supporting excerpt: None of that proves the strategy works. Closing a purchase is not the same as showing the model pays. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 260 - Selected supporting excerpt: What the deals do show is where the capital is pointed: at owning the service, not at selling it a seat. - Limits: As written in the paper. ### cl-investor-question / editorial / reviewed Label: Our reading The question for an investor in roofing AI is not whether the tools work. It is who keeps what they create. Scope: For the investor Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 262 - Selected supporting excerpt: The question for an investor in roofing AI is therefore not whether the tools work. It is who keeps what they create. - Limits: As written in the paper. ### cl-fall-line / editorial / reviewed Label: Our reading A one-task tool is a feature on someone else's roadmap. Scope: The fall Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 264 - Selected supporting excerpt: a wedge is a feature on someone else’s roadmap. Tools fall. Outcomes stay. - Limits: As written in the paper. ### cl-four-tests / editorial / reviewed Label: Our reading A point solution survives when it has something an assistant cannot copy, and four things count: data the assistants cannot get, a place in the workflow that is expensive to leave, a rule that requires it, and an outcome it stands behind. If a tool has none, the argument applies to it. If it has one or more, it does not. Scope: What would change our mind Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 268 - Selected supporting excerpt: A point solution survives when it has something an assistant cannot copy. There are four things that count. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 270 - Selected supporting excerpt: ### Data the assistants cannot get - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 274 - Selected supporting excerpt: ### A place in the workflow that is expensive to leave - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 278 - Selected supporting excerpt: ### A rule that requires it - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 282 - Selected supporting excerpt: ### An outcome it stands behind - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 286 - Selected supporting excerpt: Ask which of the four a given tool has. If the answer is none, the argument in this paper applies to it. If the answer is one or more, it does not, and we would want to hear about it. - Limits: As written in the paper. ### cl-data-owner / editorial / reviewed Label: Our reading A tool built on a dataset it owns is not a wrapper. In the Verisk and EagleView cases, the tool is the data owner's own product. Scope: What would change our mind Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 272 - Selected supporting excerpt: In our view, a tool built on a dataset it owns is not a wrapper. In these two cases, the tool is the data owner’s own product. - Limits: As written in the paper. ### cl-rules-protect / editorial / reviewed Label: Our reading Texas and Florida decide who may speak for a policyholder on a claim. Neither statute mentions software. Rules like these protect the licensed person, not the point solution that drafts the letter. Scope: What would change our mind Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. Not legal advice. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 280 - Selected supporting excerpt: Texas and Florida decide who may speak for a policyholder on a claim. Neither statute mentions software. In our reading, rules like these protect the licensed person, not the point solution that drafts the letter. - Limits: As written in the paper. ### cl-outcome-pricing / editorial / reviewed Label: Our reading A vendor paid on the result rather than the seat is aligned with its customer in a way a license cannot be. Scope: What would change our mind Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 284 - Selected supporting excerpt: A vendor paid on the result rather than the seat is aligned with its customer in a way a license cannot be. - Limits: As written in the paper. ### cl-sor-place / editorial / reviewed Label: Our reading A system of record where the owner runs the business is a place in the workflow that is expensive to leave. Scope: What would change our mind Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 276 - Selected supporting excerpt: A system of record where the owner runs the business is that kind of place. - Limits: As written in the paper. ### cl-ledger-assistants / editorial / reviewed Label: Our reading Graded supported: general-purpose assistants already do much of what point solutions sold. Scope: The paper's claim ledger Limitations: Foundation's grading against primary sources. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 347 - Selected supporting excerpt: | General-purpose assistants already do much of what point solutions sold. | Supported | - Limits: As written in the paper. ### cl-ledger-service / editorial / reviewed Label: Our reading Graded supported: becoming the service captures the budget that software never touches. General Catalyst states it as its thesis and is funding it. Scope: The paper's claim ledger Limitations: Foundation's grading against primary sources. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 355 - Selected supporting excerpt: | Becoming the service captures the budget that software never touches. | Supported | General Catalyst states it as its thesis and is funding it. | - Limits: As written in the paper. ### cl-ledger-foundation / editorial / reviewed Label: Our reading Graded partial: customers want one internal foundation, not a shelf of tools. Neither source measures what buyers want. Scope: The paper's claim ledger Limitations: Foundation's grading against primary sources. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 348 - Selected supporting excerpt: | Customers want one internal foundation, not a shelf of tools. | Partial | Neither source measures what buyers want. - Limits: As written in the paper. ### cl-ledger-obsolete / editorial / reviewed Label: Our reading Graded partial, too strong: each earlier era of AI tools is now obsolete. Plugins are a current ChatGPT feature again, and workflow automation is still sold. Scope: The paper's claim ledger Limitations: Foundation's grading against primary sources. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 359 - Selected supporting excerpt: | Each earlier era of AI tools is now obsolete. | Partial | Too strong. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 359 - Selected supporting excerpt: but plugins are a current ChatGPT feature again. - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 359 - Selected supporting excerpt: n8n still sells workflow automation as a living product. - Limits: As written in the paper. ### cl-ledger-two-percent / editorial / reviewed Label: Our reading Graded partial: software is about 2% of a law firm's revenue. Clio measures 0.58% to 1.77% of expenses, and expenses are not revenue. The order of magnitude holds. Scope: The paper's claim ledger Limitations: Foundation's grading against primary sources. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 354 - Selected supporting excerpt: | Software is about 2% of a law firm’s revenue. | Partial | Clio’s 2024 report measures software at 0.58% to 1.77% of expenses by firm size - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 354 - Selected supporting excerpt: Expenses are not revenue. The order of magnitude holds. - Limits: As written in the paper. ### cl-ledger-harvey / editorial / reviewed Label: Our reading Graded unsupported: Harvey is walking the path to becoming a law firm. No source shows Harvey operating as a law firm. Scope: The paper's claim ledger Limitations: Foundation's grading against primary sources. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 356 - Selected supporting excerpt: | Harvey is walking the path to becoming a law firm. | Unsupported | No source we opened shows Harvey operating as a law firm. - Limits: As written in the paper. ### cl-ledger-unsourced / editorial / reviewed Label: Our reading Graded unsupported: Nvidia will undercut the frontier labs with in-house hardware and compute-only pricing; venture investors are still underwriting the old playbook and do not know the ground; distribution is not a moat. No source the paper opened establishes them. Scope: The paper's claim ledger Limitations: Foundation's grading against primary sources. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 349 - Selected supporting excerpt: | Nvidia will undercut the frontier labs with in-house hardware and compute-only pricing. | Unsupported | - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 346 - Selected supporting excerpt: | Venture investors are still underwriting the old playbook and do not know the ground. | Unsupported | - Limits: As written in the paper. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 365 - Selected supporting excerpt: | Distribution is not a moat. | Unsupported | - Limits: As written in the paper. ### cl-handoffs / editorial / reviewed Label: Our reading In the point-solution pitch each step gets its own tool, each tool keeps its own copy of the customer, and the gaps between tools are handoffs. Scope: For the roofer Limitations: Illustrative, not a count of what any roofer uses. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 237 - Selected supporting excerpt: In the point-solution pitch each step gets its own tool, and each tool keeps its own copy of the customer. The gaps between tools are handoffs. - Limits: As written in the paper. ### cl-escalation / editorial / reviewed Label: Foundation position What lasts is owning the outcome: the workflows, the data they leave behind and the accountability for the result, with people handling what the system escalates. Scope: What lasts Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 39 - Selected supporting excerpt: What lasts is owning the outcome: the workflows, the data they leave behind, and the accountability for the result, with people handling what the system escalates. - Limits: As written in the paper. ### cl-compound / editorial / reviewed Label: Foundation position Three things compound when a model improves instead of replacing the product around it: the workflows a company runs, the data they leave behind and the accountability for the result. A foundation is the three together, owned by someone who answers for the outcome. Scope: What lasts Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 292 - Selected supporting excerpt: Three things compound when a model improves instead of replacing the product around it: the workflows a company runs, the data they leave behind, and the accountability for the result. A foundation is the three together, owned by someone who answers for the outcome. - Limits: As written in the paper. ### cl-people / editorial / reviewed Label: Foundation position Owning the outcome does not make the people doing the work a cost line. In a roofing company, owners and crews handle what the system escalates. Scope: What lasts Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 294 - Selected supporting excerpt: Owning the outcome does not mean the people doing the work become a cost line. In a roofing company, owners and crews are the ones who handle what the system escalates. - Limits: As written in the paper. ### cl-sliver-close / editorial / reviewed Label: Foundation position A tool earns a sliver of each job. Whoever owns the outcome answers for it, and keeps what compounds. Scope: What lasts Limitations: Foundation's reading. Foundation Projects builds a platform for roofing companies, so it is not neutral. - A shelf of tools is not a foundation (paper, Markdown edition, read 2026-10-18) / line 296 - Selected supporting excerpt: A tool earns a sliver of each job. Whoever owns the outcome answers for it, and keeps what compounds. - Limits: As written in the paper. ## Continue Start a private conversation: https://foundationproject.net/schedule Credits: Fonts: Newsreader (SIL Open Font License 1.1; Copyright 2020 The Newsreader Project Authors (http://github.com/productiontype/Newsreader)); Inter (SIL Open Font License 1.1; Copyright 2016 The Inter Project Authors (https://github.com/rsms/inter)). Music: generated with Higgsfield Sonilo. Music is optional: the page opens with a choice, "Sound on" or "Continue without sound", and says everything without it. Run: 20261018T175912157Z-ec9df8c1 Content fingerprint: 90d4e5c16b36b254419bdc51de238b9324daf349ae35d75ad917ce9bd3e043cb